Improvement
Thinkr now reads your images
Upload an image to a notebook and Thinkr transcribes its text and describes what's in it, so you can ask about the image's contents in chat — not just find it by filename.
Until now, an image you uploaded to a notebook was only findable by its filename — Thinkr stored the picture but didn't read it. So a screenshot full of useful text, or a chart with the numbers you cared about, was invisible to chat.
That changes today. When you upload an image, Thinkr reads it: it transcribes any visible text (OCR) — labels, captions, UI text, table cells, handwriting — and writes a short description of what the image shows, including charts and diagrams. That text becomes part of the source, so the assistant can answer questions about the image's contents and cite it, the same way it does for a PDF or a pasted note.
Open any image source to see what was read — the preview tells you whether its contents were indexed.
This applies to new uploads. If you have older images in a notebook, re-upload them to index their contents.